I run Bitcoin Core (with limited incoming connections) and Armory. It takes a big one-time download that took a few hours, ~50GB of disk space (with plenty free), ~650 MB of RAM out of 16GB, ~1 CPU hour out of 400 (based on current uptime and CPU time usages), and minimal bandwidth requirements. The last few of those I can pause any time I want to reclaim the extra resources.
In return, I get to use a good, secure client (Armory) that's connected to the network independent of any external service or undue reliance on peers (to tell the truth about the state of the blockchain, or to protect my privacy). And having powerful local clients, instead of overly-simplified ones, helps me learn more about the technologies behind it. I also like helping secure the network.

For me, that's an agreeable trade, so I run a full node. For some people, the requirements are relatively larger, and the rewards are less important to them, so the balance does not tip in the "run a full node" direction.

Self-interest can, in fact, be sufficient, including in my case. Altruism is a small part of why I run a full node, but is not sufficient nor necessary in my case.

And I know how average users can hurt the network. If we had fewer average users, I'm sure I could bump up my max number of connections substantially. I have to keep it low because I'll occasionally have someone want to download a huge number of blocks from me, and I have little upload bandwidth, so it interferes with anything else I'm trying to do.

Right.
If you're at another system with an online Armory installation where you could import a watch-only version of your wallet, you can view the address that way. Theoretically, an online service could implement this, but I don't know if any have: it's something of a niche thing IMO, considering how it reduces the privacy that Armory provides.

I'd suggest that instead you consider using something like Chrome Remote Desktop to let you access your computer from your phone. This would allow you to easily check the balance.

If you have multiple wallets and then keep all of the keys for them secured in the same place (whether that's a safe deposit box or the cloud), it's pointless: you should use one wallet.
Having multiple wallets allows you to do things like have an offline computer be your cold storage, as well as having a hot wallet, and have paper wallets in separate safe locations (e.g. keep 10 BTC in a safe deposit box and another 10 BTC stored as an n-of-m backup with family/friends).

At 28 bits of entropy (as a rough estimate), no it's not a good brain wallet. Given the right resources (a list of popular books and their texts in a certain edition), an attacker would only have to try 2^28 combinations to crack most brain wallets based on this scheme.
You're only short by a factor of 2^100. *cough*

I can memorize a 128-bit key fairly easily, so there's nothing stopping me from memorizing a 256-bit key except that there's no point to it: I feel quite secure with a 128-bit seed securing my wallet.

(Armory 0.92.1, Windows 7)
When you click multiple bitcoin: links (using the feature where Armory is registered to handle bitcoin: links), Armory opens them in separate windows/transactions. Since these dialogs are modal, this means that you have to send or cancel the last one you clicked before you can deal with the earlier ones (which limits your ability to copy/paste things to merge the transactions). I'd suggest that instead in that scenario, it should add all outputs to one transaction window, like if you copy each link address and "Manually Enter.." them.

I ran into this recently when I was making two purchases at once. To keep fees low, I'd rather send them both as one tx, and Armory's default behavior made that a little difficult.

http://serpco.in/#code generates keypairs for you. Send money to the addresses it generates.

You should run your memory dual-channel. Basically, if you get two of the same type of chips, they can run twice as fast. (consult your motherboard manual for details on if it supports it, where to put the chips, etc.)
E.g. I'd recommend that you get two 8GB chips (16GB total) instead of 16+4GB as you describe.

Something like this?
http://www.reddit.com/r/Bitcoin/comments/2d51b0/my_raspberry_pi_bitcoin_ticker/

No, when you send, it must be from an address that you've previously received money at. At a technical level, this is simply how Bitcoin works: not account/wallet balances, but transactions to/from addresses.

The way that Armory sends and receives bitcoins is fairly privacy-conscious, but you should still assume that someone determined could figure out which transactions are yours. If you want to obscure this to a high degree, use a bitcoin mixer of some kind or CoinJoin.
The public transaction will only show the addresses involved in the transaction, which will typically not be *all* of the addresses in your wallet.

To add some details to what the others have said about the difficulties of doing this, see https://en.bitcoin.it/wiki/Transactions
In order to create a transaction offline, you'd need to not just tell the offline wallet that you have sufficient balance, but you'd need to tell it one or more (enough to equal or be greater than your desired output, less transaction fee) transaction outputs to spend. For your typical transaction, this means:
One (or more) 256-bit txid to spend (if it's a from another transaction the offline wallet already knows about, e.g. because it was a change output, you could simply choose it from the list)
One (or more) address from your wallet (could be chosen from the list of addresses in your wallets), being the address to spend corresponding to each txid you're spending
The address(es) you wish to spend to

Transmitting all of this data manually would be tedious, to say the least.

1. Download and run Tor Browser
2. Go to https://blockchain.info/pushtx
3. ???
4. Profit! Er..Anonymity!

First of all, there are on the order of 500 transactions in a typical block (currently at https://blockchain.info/), so "hundreds of transactions" is not nearly old enough - maybe instead, you can store the balance of ~3 million (current) addresses with their balances in a ledger.
Second, it reduces the security. E.g. what if someone created a ledger (via 51% attack) that says that they have every bitcoin in circulation? How would we know that's the true result of the chain of transactions, and not fictional? The cryptographic signatures authorizing each transaction are the only thing preventing a 51% attack from stealing all bitcoins (instead of just doing other disruptive things).

When I realized that poll options are being added from comments on the thread, this thread lost all semblance of sanity. Or, it tries to represent the ever-growing blockchain as an ever-growing poll. New thread idea: What happens when the poll in this thread gets too big?

This wallet is pretty mediocre compared to the Web version (probably OS X too): this is just Andreas's app with add-ons added on.
Is it planned for this wallet to be deterministic/passphrase-based?

Here's a suggestion: randomly generate a 32-bit unique identifier, store it in the Armory config files, and report that. It's not perfect, since someone with multiple folders will be counted multiple times, but it's pretty good.

We're not entirely in tin foil hat land, because the whole point is that you have multiple papers, and only load one at a time. Let's say each paper is worth 0.1 BTC, and you carry 10 papers at any one time. Typically, you'll only load one paper at a time, so the most that can be stolen is about 0.2 BTC. I think it's fair to assume that if your coins do get stolen (because your phone can no longer be trusted), the losses are limited to 0.2 BTC, and you can simply not scan any more papers in.

Not quite. With his plan, someone has to steal both his (physical) wallet and his phone (and then spend/transfer it before the rightful owner realizes the theft and transfers the coins elsewhere) in order to steal his coins. This is of limited use since the wallet and phone are typically carried on your person, but at least it protects against digital-only attacks, and some in-person theft scenarios.

I'd add that you should encrypt the transaction with a passphrase/password before writing it to the paper. Unfortunately, this further complicates the retrieval process: you need a decryption app on the phone. But it does mean that the paper is useless without something that only exists in your head. In this case, you could even store the TXs digitally on your phone and/or the Internet, since there is no longer a security concern.

You're wrong. Security through obscurity is not secure. You should assume that the enemy knows everything about your system except your private key.

Actually, he probably knows where the valuables are kept, because the average person is stupid and predictable, even when they try to be clever (this is true of hiding places for valuables, and doubly so for passwords). It'd be much better if you could instead stick an unbreakable safe in the middle of your living room.

No, of course not.

Credit card adoption didn't require that retailers stop accepting cash. People can simply see how much better it is (don't have to deal with change or writing checks, buy now pay later, rewards for using), and transition on their own. Once retailers start passing on the savings they get, people will see how much better Bitcoin is than credit cards, and transition.

From http://webbtc.com/stats, about 99.10% of outputs are to a pubkey hash. Another 0.77% are to pubkeys. About 0.13% are others: multisigs, P2SH, OP_RETURN, and non-standard unknown scripts.

For the sake of a rough/Fermi estimate: all outputs are pubkey hash.